Kristian Agasøster Haaga
Kristian Agasøster Haaga
> We test one case with 2 and one with 3 input arguments, which is the only thing that changes in the source code when the functionality is applied. Also,...
> Hello, I'm interested in taking on this issue and optimizing the AAPE function to improve its performance. Hey, @EidanGar! We're happy to hear that! > Before I claim the...
> Are datasets significantly different from Vectors that they cause 4 more allocations? @Chronum94 In the applications we have here, `AAPE` would typically be called with *elements* of `Dataset`s, which...
Ah, sorry, my bad. I didn't test the code. Broadcasting like that isn't implemented `Dataset` yet. You'll have to do ```julia import Entropies.AAPE # make sure you've checked out main...
Hm. Not entirely sure about this. Should we instead *require* the input to be sortable (i.e. the user has to map their data to some sortable format)? That would make...
> I think in the future there should be a version implemented for non-sortable collections. But this is extremely low priority because a user can always map their vector to...
> We could check something like > > ``` > julia> hasmethod(isless, (T, T)) > false > ``` > > with `T` the element type of the vector. Yes, this...
In the current API, the docstring of `outcome_space` is ```julia """ outcome_space(o::OutcomeSpace, x) → Ω Return a sorted container containing all _possible_ outcomes of `o` for input `x`. For some...
I haven't done any sort of testing for speed etc, but an obvious (and perhaps the easiest) solution to this issue is precisely what @Datseris says above - convert to...
In any case, anyone working on this should wait until #347 is merged.